try disconnecting the 2 ribbon cables for the soundboard on the top of the MPU

http://www.techdose.com/projects/Stern-Pinball-SB100-Sound-Board/346/page4.html

hope this works & can move on from there.

Have you confirmed if you are getting address and data line activity to the ROMs? Are the Chip Enable pins on the ROMs pulsing?

The 6800 reset vector table is held in the ROM at U6. It points to code starting in the ROM at U5. The board must be able to effectively access those two ROMs to begin the power on self test which first switches the MPU LED off then starts performing the ROM checksum test.

Can you post some clear high resolution pictures of the board front and back - preferably taken outdoors during daylight in the shade so your camera can achieve the best result?

If you take U11 out, the Led will never turn off.... it's connected to U11!

Leave the 6800 in with nothing else, power it up, use a logic probe you should see address activity on all the address lines as the processor runs through its full address range, with each successive address line being 1/2 as fast as the previous.

If you get that, you know the processor is doing something and is probably ok. If it's not doing it, short pins 39 and 40 on the 6800 momentarily to force a reset, then see. If it works then, you have a reset circuit issue. If it's not working, I'd suspect socket for u9 or the processor itself, or something REALLY off with the reset circuit (repeating for some reason?)

Next put in U10, U11, and the leon test chip. The test chip should make the LED blink on and off (I forget if the test chip needs the 6810 ram, it might.... if it does put that in as well). You can also check all the outputs of the PIA chips u10 and u11 with your probe at this point, they should be pulsing.

If you get to this point, likely the board will boot if you put good eproms in. Make sure your eprom jumpers are correct.

Yes, you can certainly check for pulsing on address pins using your scope. The LED thing you made is sort of a logic probe anyway just a little harder to use.

AMI branded chips fail more often than motorolas. so it may very well be the 6800. Do you have any others you can try?

If it's a dual or quad trace scope you can get really fancy and actually see the differences on the address lines. Hook one channel up to A0 and the other channels to A1, A2, A3. A3 should be twice as slow as A2, which is twice as slow as A1, which is twice as slow as A0.

You'll just see pulsing on it, nothing special. If you see a flat signal.... well.... your 6800 is dead.

The VMA (Valid Memory Address) signal goes to a high state when the CPU places a valid address onto the bus to access external devices (ROMs/RAMs/PIAs) etc. When the CPU is internally executing instructions, VMA will go low. In other words, the VMA signal will constantly have activity between high and low.
The 2.8 volt specification is a poor mans way of checking if the signal is active using a DC volt meter, i.e. sometimes the signal will be high (5 volts) sometimes it will be low (0 volts) so a DC volt meter will just give you an average that's somewhere in the middle. If the signal is stuck on high or low, that's when there's a problem with it.

Now what about at the ROM chips, do you see activity on all the ROM pins which are not connected to ground or 5 volts?
How about pin 34 on the CPU which is the R/W (Read/Write) signal? I had a faulty CPU (AMI) that couldn't drive the R/W signal.

Your board looks in very nice condition. As slochar mentioned, those AMI branded parts have a high failure rate so it's probably your first suspect.

BTW, can you rotate your pictures left 90 degrees, I'm getting a sore neck

Pinside_RotateImage.gif

Great!
Temporarily connect a jumper wire from U12 pin 3 to U14 pin 7 and you should get the 7th LED flash on power up.

I always wonder what the heck makes the AMI chips so unreliable, they must have a consistent design flaw to them as the failure rates is well beyound STM, Moto, and Hitachi.

I would hazard a guess that they passed more "marginal" wafer that the other makers rejected instead to save fabrication costs. If the chips worked for the 5 year design life, they probably considered that good enough.

I would guess it's more to do with some specific materials AMI used in the fabrication of their parts that broke down for whatever reason, oxidization, internal current leaking, unexpected reaction between materials over long term use, etc.
It's happening in all their parts used on these boards, CPU, PIAs and RAM chips.

The plastic packaging on your AMI CPU with the logo cutout and rectangular end cutout is typical of their parts that fail a lot, your original AMI 6810 and 6821 look like they came from a different fabrication plant with a generic plastic package - maybe those were licensed from someone else. I don't remember if I've come across those - not any that have failed anyway.